Ross Vandegrift wrote: >>Yeah, especially since processors are so slow and expensive >>nowadays. >> >>There are many people who don't need high data rates, but >>value data integrity. > > Is there any case where RAID10 wouldn't provide better redundancy? Of course. Raid6 would protect against ANY two disk failure, while you CAN think of a two-disk failure where RAID10 goes belly-up. And it sure is a good thing to KNOW you have 100% redundancy left when one disk gives up. On the other hand, a cleverly distributed RAID10 can guard against a controller failure in a 2-controller setup ; a RAID6 of more than 4 disks can't unless you use a maximum of 2 disks per controller. Of course you can also think of multi-disk failures up to half the total number of disks which RAID10 would survive, but the point is, RAID6 -guarantees- protection against a two disk failure, something RAID10 can't, at the cost of only N+2 disks, not N*2. So I definitely see a point in implementing RAID6. The beauty of this idea is that you can extend it to guarantee protection against any number of multi-disk failures you want. So if you only have two controllers and want to protect against controller failure as well as guarantee protection against a two-disk failure, there is no other way than RAID6+ at N*2 disks. Of course those parity stripes will load the processor some, but that shouldn't be a problem on current hardware. Other than that, write performance shouldn't be much worse than RAID10. How about not adding a RAID6 driver, but extending the RAID5 driver some to make the number of parity stripes a variable? I don't know how much extension that would require - perhaps it would bog down the RAID5 driver too much in single stripe use, so we still might be better off with a separate RAID6(+) driver. But it looks like most of the code could simply be copied. Marcel - To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
